CHAPTER 13
Historical Data on Species Occurrence
BRIDGING THE PAST TO THE FUTURE
Morgan W. Tingley
Abstract. Understanding climate change impacts on the environment necessitates adapting monitoring strategies to evaluate long-term changes on the order of decades and centuries. Historical data, such as museum specimens, original field surveys, and other records, provide one potential source of information that can provide inference on changes over long periods of time. Historical data are often undervalued, however, as these data were typically collected for other purposes and are seen as difficult to integrate with current data collection efforts. Here, I describe sources of historical data and key considerations for using them, and give examples of how these diverse sources are informing our current understanding of how species respond to change.
Key Points
• Historical data are increasingly used to understand how climate change is currently affecting biological resources.
• Museum specimens, atlas data, old surveys and studies, and field notes are all rich sources of occurrence data that can be used with modern comparisons.
• Differences in methods and other data-specific traits can create challenges for making valid comparisons, yet strategies exist to overcome many issues.
• Integrating historical data into ongoing climate change monitoring can benefit resource managers.
UNDERSTANDING THE FUTURE BY LOOKING TO THE PAST
Climate change during the 21st century is expected to impact species in a multitude of ways, some of which we can predict, but many of which are unknown. Based on what scientists have already observed, phenological mismatches, range shifts, and both local and regional colonizations and extinctions are occurring currently, and are expected to increase both globally (Root et al. 2003, Chen et al. 2011, Anderegg and Root, this volume) and in California (Moritz et al. 2008, Forister et al. 2010, Tingley et al. 2012). Climate change holds consequences for managed systems, as the biological value of these systems may be affected by shuffling species assemblages and modified biological processes. In addition to working toward reducing the rate of climate change, our ability to protect our natural resources may be significantly aided by investment in inferential methods that provide understanding of how these resources are likely to change in response to ongoing climatic processes.
The premise behind this work is simple: We have the opportunity to improve our understanding of what could happen in the future by exploring what has already occurred. Understanding species occurrences in the recent past allows the development of a strong baseline to measure both current occurrence patterns and judge future changes. Since climate change impacts are generally long-term trends that can be obscured through short-term variation (stochasticity), broad temporal ranges in baseline data provide greater inference and perspective on changes. Particularly if past data are rich across a range of time periods, it is possible to separate short-term stochasticity and natural variation in species’ occurrences (e.g., due to resource cycles) from long-term trends associated with shifting environmental gradients.
Work on the American pika (Ochotona princeps) provides a highly relevant example. Historical data (in this case, the collection of museum specimens) from various isolated mountaintop populations throughout the pika’s range have provided a comparative baseline for current surveys at these same locations (Beever et al. 2003, Erb et al. 2011). Comparing modern pika presences to areas of known historical occurrence, researchers have been able to document local extinctions over the 20th century. Importantly for conservation, different key factors (summer temperatures, drought stress) explain these extirpations in different locations. This is just one example of the many studies that have used historical data to under-stand climate change impacts in western North America (Table 13.1).
The greatest stumbling block to understanding changes in the recent past in a particular landscape is simply the availability of data. Although they are not always available, baseline data can come from many sources, and the original purpose of the data need not be the same as the purpose now. The specific purpose for collection can bias contemporary comparisons, however, through determining how data were collected (e.g., random sampling) or which aspects of data (e.g., habitat) were collected. Additionally, all occurrence datasets (i.e., data that pinpoint a species to a location and a time) can include false presences (i.e., misidentifications) and false absences (i.e., failure to detect a species where it occurs) that can bias comparisons. Differentiating false presences and absences from true presences and absences is an important analytical step. Thus, to unlock the value of historical data, we need to address two questions. First, what are potential sources of historical data? Second, how can historical data be compared in an unbiased way with recently collected data?
FINDING HISTORICAL BASELINE DATA
The first challenge in establishing a historical baseline for species occurrences in an area is simply uncovering historical data. “Historical” in this context does not mean preindustrial, nor does it necessarily mean a century ago, as rich data sources from those periods rarely exist. While truly historical examples do exist [such as Henry David Thoreau’s notes on flowering times at Walden Pond from the mid-19th century (Primack et al. 2009)], historical is used here to denote a time period that is relevant to climate change. Ideally, “historical” data provide a point of reference with which a temporal comparison to the present will be meaningful. Depending on the rate of climate change in a region, the extent of time between past and present necessary for a meaningful temporal comparison can vary. Throughout much of North America, anthropogenic climatic warming can be detected as early as 1900, but the recent trend of accelerating warming is not obvious until 1960, with the greatest changes since the late 1970s (Mastrandrea and Anderegg, this volume). In the present context, historical occurrence data provide perspective for modern distributions, so the time frame that defines “historical” may vary by the movement and dispersal abilities of focal taxa. For example, data from 1980 would provide very little utility if one is comparing distributions of long-lived canopy trees, yet one might potentially detect strong changes if comparing flying insects.
TABLE 13.1
Published examples, 1990–2014, from the western United States of historical data that have been compared to contemporary data in order to inform on the effects of different temporal drivers of change
INFORMATION BUILDS ON TABLE WITHIN TINGLEY AND BEISSINGER 2009
1 The number of species chosen for temporal comparison within the study. Infinity (∞) implies that all species (detected and undetected) within the community were studied.
2 The number of sites implies the number of sampling plots—whether lakes, forests, quadrats, or point counts—that are matched across temporal resurveys. The scale of individual sites depends on the resolution of historical data. Original survey data often have smaller discrete sampling units, allowing more detailed and more numerous resurvey samples.
3 The specific temporal drivers of species change that were explored or tested within the study by comparing historical data to contemporary data. Here, the term “land-use change” has multiple meanings, including: changes in land tenure, landscape, or habitat conversion, and changes in management (e.g., grazing). 4 The maximum time span of comparison from the earliest historical datum to the most recent contemporary datum.
The most useful historical data contain three components: A taxonomic identification, a geographic specification, and a date. These three components define a “collecting event” for all occurrence data. Identification challenges vary by data source and taxa; although misidentifications of species exist within natural history collections (Miller et al. 2007, Graham et al. 2008), physical specimens (and sometimes photographs) provide a means to reverify unusual occurrence records. With survey data, either trust has to be placed in the skills of the original observer or unusual records need to be sufficiently vetted using expert knowledge. The second component, geographic location, may not be precisely specified, and thus occurrence locations frequently have an associated uncertainty or error. Generally, the risks of misleading conclusions due to geographical uncertainty are greater in areas of increased habitat or topographical variability (Rowe and Lidgard 2009). For example, if it is unknown whether a sighting was at 500 or 1500 m elevation, then it may not hold much value for measuring climate-induced elevational range shifts. Similar to location uncertainty, uncertainty in the date of an occurrence record may affect the utility of that record in a comparative analysis, such as a phenological analysis of seasonal change.
Occurrence data with these three components can come from a wide variety of sources. For many long-established managed areas (e.g., national parks and forests in the United States), routine species monitoring likely extends prior to 1980. These original surveys have the potential to provide high-resolution information that can be resurveyed at a fine scale, for example, the revisiting of historical American Marten census plots at Sagehen Experimental Forest in northeastern California (Moriarty et al. 2011). While unit-specific original surveys are a great source when available, occurrence data can also be held by outside sources as part of data collected for a different purpose or part of a larger survey effort. A wide variety of sources provide this historical occurrence baseline, including biotic atlases, field notes, published literature, and museum specimens.
Combining herbaria, invertebrate, and vertebrate collections, natural history museum specimens are an incredibly rich source of occurrence data, on the order of billions of independent specimens collected (Graham et al. 2004). While museum-based specimen collecting continues today throughout the world, specimen collecting is closely associated with geographic exploration and biodiversity cataloguing, and thus particularly in North America, most specimens are historical in nature, although this differs by taxa. In California, bird specimens were predominantly collected in the first half of the 20th century, while amphibian and reptile specimen collecting peaked in California around 1970 (Figure 13.1).
Historically, specimen collectors were motivated by a desire to create collections that were physical embodiments of nature’s diversity (Box 13.1). By the early 20th century, Darwin’s lessons on the sources of natural variation were well known, and western naturalists sought to maximize the taxonomic and phenotypic diversity of collections through broadening their geographic coverage. As a result, this period was a golden age in many zoological fields for describing geographical variation in taxonomy— and these discoveries were backed up with physical specimens and often further described in field notes. Collection efforts in the second half of the 20th century shifted toward documenting morphological variation within species, resulting in large specimen “series” of the same species. The cumulative result of 20th century scientific efforts is an extensive geographic coverage of museum specimens, including specimens from both public and private lands (Figure 13.2). Online databases combining specimen records from tens to hundreds of museums around the world now exist for most taxa, with prominent examples including the Global Biodiversity Information Facility (GBIF: http://www.gbif.org/), Map of Life (http://www.mol.org), and VertNet (http://vertnet.org/). VertNet works as a data portal, patching together taxa-specific occurrence data for birds, mammals, reptiles, amphibians, and fish. These databases have dramatically increased access to museum-vouchered occurrence data.
FIGURE 13.1: Temporal variation in the number of avian and herpetological specimens collected in California since 1850 and currently held in North American natural history museums. Bird-specimen collection was strongest during the first half of the 20th century as the natural history of California was being “explored.” Collection fell off after the Second World War as collection efforts, when made, were focused internationally. Reptile and amphibian collections grew steadily through the 20th century with a drastic rise after 1950. This correlates with a reevaluation of morphology and taxonomy that resulted in the discovery of numerous cryptic species. Source: Data downloaded from VertNet.
Field notes are an additional notable component to documenting historical species occurrences. While specimen records provide quick access to species occurrences, they often provide very little associated information, called “metadata,” which provides context and background. In contrast, field notes made during specimen collecting expeditions or during other natural history and ecological survey work often provide a wide variety of useful metadata in addition to the essential collecting event for an occurrence. These metadata can provide more exact temporal and geographic precision (e.g., “A quarter till 8 we stopped along the west side of Long Meadow, 500 ft north of the stream outlet”), as well as other useful details, such as weather, survey effort, collecting method, ecological context, and behavior. While field notes are often the only source of occurrence data for observations, the metadata gathered from field notes can also be applied to associated specimen records, thus greatly increasing their analytical value (e.g., Moritz et al. 2008). While not all field notes are properly archived and made accessible, several field note repositories have recently begun the process of digitization and providing online access; prime examples include the Museum of Vertebrate Zoology at UC Berkeley (http://bscit.berkeley.edu/mvz/volumes.html) and the Smithsonian National Museum of Natural History (http://www.mnh.si.edu/rc/fieldbooks/index.html). Even in electronic form, field notes often require extensive research in order to pull out occurrence records and associated metadata. The reward can be proportional to the effort, however; field notes have been used successfully to understand long-term changes in occurrences and populations of a diverse group of species, from birds (Colwell et al. 2002, Martin et al. 2004) to fish (Labay et al. 2011) to invertebrates (Daniels 1998, Blalock-Herod et al. 2002) to plants (Brusca et al. 2013).
Historical occurrence data may not be the only historical data that will be important when analyzing long-term changes at one location, as we are often interested in relating occurrence changes of organisms to ongoing processes such as climate, land-cover, and land-use change. Historical climate layers are generally derived from long-term continuous weather monitoring stations. In the United States, data from these weather monitoring stations are aggregated and curated by the United States Historical Climatology Network (USHCN: http://cdiac.ornl.gov/epubs/ndp/ushcn/ush cn.html). They provide access to individual daily and monthly weather records for over 1200 unique stations with some records beginning as far back as 1895, although data quality from individual stations may deserve close attention. If continuous climate coverages are needed, there are several options. The Climatic Research Unit (CRU; http://www.cru.uea.ac.uk/) provides monthly and yearly climatic surfaces for the entire world (at large 5° × 5° grid cells) going back to 1850. Within the continental United States, the PRISM Climate Group (http://www.prism.oregonstate.edu/) produces monthly grids going back to 1890 at a much finer cell size (800 × 800 m2). Compared to climate data, historical land-use and land-cover data can be harder to find as they require the prior existence of detailed—often, hand-drawn—maps. The USGS provides access to resources that provide land-cover data from multiple time periods (http://landcover.usgs.gov/landcoverdata.php), but specific data availability may differ by state and locality. In California, for instance, the Wieslander Vegetation Type Mapping Project (VTM: http://vtm.berkeley.edu/) is a highly informative dataset providing information on changes in forest composition and land cover in many parts of the state since the 1930s.
FIGURE 13.2: Spatial variation in the availability of historical (pre-1980) avian and herpetological specimen-based occurrence data in California with respect to land ownership. Image shows California divided into 12 × 12 arcmin grid cells, colored according to the total number of unique geographic sites with bird-, reptile-, or amphibian-specimen data. Each unique site represents a place visited by museum collectors where information on one or more species is contained within the specimen record, and also likely within field notes. Grid cells with no specimen locations are not colored. The overall coverage shows the wide availability of species information from the specimen record, including areas currently predominated by both public and private land ownership. Note that this map is restricted to showing digitized and georeferenced records of specimens held in North American museums and is thus not a complete map of the true existing coverage of museum data for these taxa. Source: Map pools data for specimen records downloaded from VertNet.
BOX 13.1 • Joseph Grinnell and origins of systematic occurrence data
The flora and fauna of California and the western United States had already been well documented before Joseph Grinnell was named the founding director of the Museum of Vertebrate Zoology at the University of California, Berkeley, in 1908, but Grinnell ultimately changed the nature of the collection of natural history information. Prior to Grinnell, there was little effort invested in planning how specimen collections should be built. Generally, diversity was maximized, but additional information was rarely collected. Grinnell saw the value in metadata, as transcribed via field notes:
Our field-records will be perhaps the most valuable of all our results. . . . any and all (as many as you have time to record) items are liable to be just what will provide the information wanted. You can’t tell in advance which observations will prove valuable. Do record them all!
(Field notes of J. Grinnell, 1908)
Grinnell had the foresight to directly link the systematic collection of specimens and their metadata to the assemblage of information on occurrence and ecology:
It will be observed, then, that our efforts are not merely to accumulate as great a mass of animal remains as possible. On the contrary, we are expending even more time than would be required for the collection of the specimens alone, in rendering what we do obtain as permanently valuable as we know how, to the ecologist as well as to the systematist. It is quite probable that the facts of distribution, life history, and economic status may finally prove to be of more far-reaching value, than whatever information is obtainable exclusively from the specimens themselves.
(Grinnell 1910)
Grinnell expounded on his new method of collecting field notes when he could, trying to convince others to adopt the method:
These notes were written ‘on the spot’ from time to time during the three or four hours of observation. They show the nature of a certain type of field observations, how these may be recorded in a running narrative style, and there is perhaps some information presented of general interest to the student of living birds.
(Grinnell 1912)
Grinnell ended up succeeding in his efforts. His systematic and devoted collection of field notes became known as the “Grinnell Method” and greatly changed the means by which natural history information has been collected.
LITERATURE CITED
Grinnell, J. 1910. The uses and methods of a research museum. Popular Science Monthly 77:163–169.
Grinnell, J. 1912. An afternoon’s field notes. Condor 14(3):104–107.
HISTORICAL RESURVEYS IN THE WESTERN UNITED STATES
The availability of historical data provides an opportunity to conduct a resurvey, where historical data are compared with newly collected data. Researchers are increasingly gaining inference on long-term changes in species distributions and abundances through resurveying locations with historical occurrence data (Shaffer et al. 1998, Tingley and Beissinger 2009). A review of published resurvey studies can help illustrate opportunities for learning from the past at both local and regional scales. In the western United States alone, there have been at least 30 published studies since 2000 involving resurveying for taxa using historical data (Table 13.1). These studies illustrate the diversity of potential inference available from historical data and can be lumped into three groups: Extinction and colonization, range change, and abundance and community composition.
Studies of extinction and colonization tend to focus on small numbers of species of conservation concern (fewer than 10 species), and use historical occurrence data to pinpoint where species used to be, and compare that to current, higher-resolution data. This type of inference has played a key role in documenting amphibian declines over the 20th century. For example, Davidson et al. (2001) explored extirpations of the California red-legged frog (Rana aurora draytonii) by determining the current population status of 237 sites of historic occurrence. Looking at the pattern of decline helped Davidson et al. to evaluate different causal hypotheses, including climate change, habitat loss, pesticides, and UV-B radiation.
To study climate change impacts in particular, researchers have been interested in using historical data to help describe changes in range (Table 13.1). This type of inference builds on extinction and colonization studies by using historical data to delineate species’ distributions across gradients, whether latitudinal, elevational, climatic, or other. In order to make robust inferences on range change, these studies often take advantage of historical surveys that were repeated in the same place (see “strategies for using historical data”). Additionally, while inference on range change can be made from specimen occurrence data, field notes are often necessary in order to learn where species did not occur historically. For example, in order to understand altitudinal shifts in small mammals of the Ruby Mountains in Nevada, Rowe et al. (2010) supplemented specimen data with field notes to determine non-detections.
Studies focused on changes in species abundance and community composition (Table 13.1) require even more detailed historical data. To meet these information needs, abundance and community studies predominantly use original survey data. For example, Lutz et al. (2009) used data from historical vegetation plots in Yosemite National Park to examine declines in large-diameter trees. As they were unable to relocate the original plots from the 1930s, Lutz et al. compared the historical surveys to two modern surveys and attempted to control statistically for sampling bias. In this example, information content and data quality of historical data made up for highly uncertain geographical location.
Comparing resurvey studies in the western United States since 2000, several trends are evident. First, there is a negative correlation between the time span of inference and the average information quality of historical data. Extinction and colonization studies have the largest average time span of comparison (Table 13.1), but these studies also generally make the simplest types of comparisons. To make unbiased inference on range change or changes in abundance or community structure, one needs more and more detailed data. While one can go farther back in time with the specimen record, greater detail is often contained in more recent survey work. This trade-off in information content does not necessarily mean a trade-off in inference on drivers of change. On average, extinction and colonization studies examine just as many drivers (e.g., land-use change or climate) to explain temporal differences as other studies. In fact, range change studies generally examine fewer drivers despite focusing on more species and requiring broader sampling. This limited focus is probably due to the recent focus on climate change as a driver of range shifts, whereas other drivers have not historically been associated with these shifts.
STRATEGIES FOR USING HISTORICAL DATA
Historical data are exciting because of their untapped potential, but by nature, historical data have extra uncertainties that can bias comparisons to contemporary data. Consequently, comparing historical data to modern data in a manner that limits the potential for bias is an important concern. The best strategies for addressing bias depend on the form of the assembled historical dataset.
The most basic form of historical data is a semi-random collection of species and their collecting events, with only broad spatial or temporal resolution. These “presence-only” data lend themselves to the development of baseline species lists, but must be used with the explicit understanding that the lists are incomplete because an unknown number of additional species may have been present at the site but not sampled. Consequently, presence-only data are best used for inference on whether a baseline species has been extirpated in an area. While false absences are unknowable solely from presence-only data, there are some ways to get around this limitation. For example, other studies that document how detectability differs by species and habitats may, if applied judiciously, provide transferable inference on the severity of false absence biases with a given dataset. Building distributional models off presence-only data may also provide a means to assess whether species occur in an area (e.g., Comte and Grenouillet 2013).
Information content can be added through field notes and other sources of metadata to increase the utility of presence-only data. A valuable piece of information is whether occurrences were collected as parts of unique-survey events where all species detected were recorded. For example, a researcher may have gone hunting one morning and returned with specimens of six different species. It is possible that she detected other species while out but did not collect them. Field notes could fill in these data gaps, informing whether any additional species were detected, and if so, the identity of those uncollected species. This would then allow the bundling of all species identifications into a single collecting event where it is known both what was detected and what was not detected. At this point, the data have gained information content and become “presence and non-detection” data (Tingley and Beissinger 2009). Through knowing both what was detected and what was not detected, estimating colonization becomes possible.
Presence and non-detection data, however, suffer from the hazards of false absences. The problem is that the presence of something is ultimately knowable (if it is seen, it is present), yet the absence of something is rarely conclusive (if it is not seen, it may still be present). Since estimating colonization or extinction relies on a change in status from absent to present, or vice versa, false absences can lead to systematic bias by artificially inflating both colonization and extinction estimates. For presence and non-detection data, this bias is likely to be high for rare or cryptic species but may be negligible for always abundant or very visible species. On the other hand, ubiquitous species may be left out of nonsystematic survey records, as many observers are more focused on finding rare or target species. Comparing contemporary data to individual historical surveys may result in biased conclusions due to false absences.
Additional information can be added to occurrence data if discrete surveys are temporally replicated within a relatively short period of time at a location. For example, if a researcher went out collecting three days in a row at the same location and kept notes of what she found each day, the repeat surveys can be used to estimate detectability. Detectability, or the probability of detecting a species given its presence at a site, provides a means to reevaluate apparently vacant sites and ask whether a species was truly absent from the site or simply present and non-detected. Only through estimating the probability of true absence of a species from sites can occurrence data provide unbiased inference on both colonization and extinction of species over time.
Quantitative ecologists have arrived at a variety of statistical ways to estimate the true probability of presence or absence of a species, but one way, known as occupancy modeling (MacKenzie et al. 2006), has gained acceptance in recent years. Mackenzie et al. (2006) provide a full treatise on how to build and use occupancy models, while Tingley and Beissinger (2009) provide specific advice for using occupancy models with historical data. The great strength of occupancy modeling lies in both its power to provide unbiased estimates of an ecological state of great interest (i.e., the true absence of a species) and its flexibility to adapt to different data formats and questions. While they can be complex, occupancy models constitute an analytical tool capable of addressing the biggest challenge of using historical data—the task of differentiating between true occurrence changes and false changes that appear due to different methods, purposes, and changing detection probabilities over time.
INCORPORATING HISTORICAL DATA INTO CONTEMPORARY MONITORING
Historical occurrence data can play only a minimal role in management and conservation unless we are able to compare it to contemporary data. A challenge for land managers and others engaged in monitoring is in adapting existing data collection schemes so that data can be used to look both backward and forward. Implementing changes to any monitoring scheme will necessarily be site and program specific and may include numerous other considerations; but integrating historical data into a program for the purpose of monitoring contemporary change can be a gradual, multistep process that need not initially affect current monitoring. As a resource to managers, the following steps are presented as general guidelines.
STEP 1. BEGIN BY DOCUMENTING PAST OCCURRENCE Data on past occurrences have multiple uses, and provide a general baseline for assessing future occurrence changes. Many natural areas have compiled or partially compiled documentation of species occurrence over time. Pulling these information sources together opens the door for further exploration of historical data.
STEP 2. CLASSIFY DATA STRUCTURE What is the format of the historical occurrence data? What kinds of metadata are available? The answers to these questions will help you to classify data by information content, such as “presence-only” or “presence and estimable absence.”
STEP 3. COMPARE DATA STRUCTURES How does the historical data structure compare with data being collected now? Assess the opportunities for determining colonization, extinction, range shift, community composition, and abundance.
STEP 4. DEFINE AND—WHERE POSSIBLE—FILL DATA GAPS Based on how historical and contemporary data differ, identify areas where inference can be improved through the filling of data gaps. For example, what data would be needed to upgrade from presence-only data to presence and non-detection data? Both historical and contemporary data can be enriched.
STEP 5. CONDUCT COMPARISON When both data sets have been organized and evaluated, a comparison between historical and contemporary data can be conducted. Conducting a comparison may necessitate normalization of data by survey effort or other factors, or incorporation of these factors into analytical methods. The results of this comparison can be used to prioritize future monitoring as well as to generate new hypotheses that can be evaluated using historical data.
STEP 6. REPEAT AND REFINE During the process of the first five steps, additional questions may develop, new data sources may be identified, or sampling frames may need to be expanded. All of these are valid reasons for repeating the process and incorporating additional data so as to refine the comparative process.
STEP 7. EXPAND THROUGH COLLABORATION Historical–contemporary comparisons at single locations may be more informative if they are expanded to a larger spatial perspective. To this end, single management units may ultimately benefit from working with other managed systems or collaborating with academic researchers to expand their comparison outward.
STEP 8. ARCHIVE AND SHARE COMPARISON If collaboration (step 7) is not feasible or logical, local studies of temporal change can be extremely valuable when incorporated as part of larger regional or continental meta-analyses. Single-unit studies can greatly benefit the broader understanding of long-term temporal processes just by archiving data in publicly accessible formats (see Box 13.2) and distributing or publishing findings.
BOX 13.2 • Contributing to Occurrence Databases
Looking forward, the ability of scientists to understand how species and systems are responding to climate change may depend on the availability of data that are broadly distributed through both time and geographic space. This broad type of data cannot be collected by scientists alone, but relies on the efforts of hundreds or thousands of “citizen scientists.” The datasets and websites below represent both sources of information and places to share observations collected in many current-monitoring programs.
Phenological Data
in the United States, the aggregation of phenological data is broadly overseen by the US National Phenological Network (USNPN: http://www.usanpn.org/). Within the USNPN, there are regional projects with more narrow spatial focuses that aim to build local capacity for phenological monitoring. For example, in 2010 the California Phenology Project (http://www.usanpn.org/cpp/) was launched to focus on building phenological monitoring in the state.
Occurrence Data
the systematic collection of species observations from scientists and citizen scientists is rapidly growing in capacity. Many states have online portals for the collection of information on rare or threatened species observations (e.g., California Natural Diversity Database: http://www.dfg.ca.gov/biogeodata/cnddb/). Taking advantage of the broad appeal of bird-watching, Cornell University and the National Audubon Society launched eBird (http://ebird.org) in 2002, which accepts data in a variety of formats. Since then, eBird has expanded to accept bird sightings from anywhere in the world and scientists are actively using data to monitor bird populations and phenology in real time. As of 2014, eBird only accepts bird records, but the idea is starting to expand, as exemplified by the North American “Butterflies I’ve Seen” database (http://www.nababis.org/) and the alltaxa iNaturalist project (http://www.inaturalist.org/). iNaturalist aims to collect observational data with the same level of information content as museum specimen data, including novel metadata such as photo vouchers and geographic uncertainty, with data quality assessment built-in.
OPPORTUNITIES TO LOOK TO THE FUTURE
Joseph Grinnell, lauded Western naturalist and advocate for extensive sampling and note-taking, commented in Popular Science Monthly in 1910:
At this point I wish to emphasize what I believe will ultimately prove to be the greatest purpose of our museum. This value will not, however, be realized until the lapse of many years, possibly a century, assuming that our material is safely preserved. And this is that the student of the future will have access to the original record of faunal conditions in California and the west, wherever we now work. He will know the proportional constituency of our faunae by species, the relative numbers of each species and the extent of the ranges of species as they exist to-day.
Grinnell was tackling a problem that many managers face: How do we know what information will be the most useful in the future? Grinnell’s solution was to collect as much information as possible because one cannot predict the future. While Grinnell’s broad intensity and effort paid off (e.g., Moritz et al. 2008), the variety, detail, and types of data that can be collected today are vastly greater than in Grinnell’s day, making his strategy impractical. Resource managers and conservation professionals today face a greater number of choices in monitoring schemes, yet as data are used in ever more ways, the trade-offs for choices are becoming less clear. This chapter has focused on using historical data, but today’s data will all be “historical” soon. Thus, while we do not know exactly what data will be most useful in the future, we are at a good point to look to the past and see what lessons may be used to inform future decisions.
One lesson from the past is the importance of repeatability. While arduously collecting extensive species data can have valuable applications, contemporary data in the future may be valued by how easily comparative data can be collected. For example, Shavit and Griesemer (2009) explore the ambiguities in defining a resurveyable “locality” from Joseph Grinnell’s specimens and field notes, noting that even specimens with assigned latitudes and longitudes may not be specific enough to confidently revisit the exact same spot. While dealing with uncertainty and ambiguity is part of why we use specialized tools for dealing with historical data, to the extent that we can make this process easier for ourselves or others in the future may be worth the effort. Repeatability, subsequently, comes from collecting metadata at multiple scales and resolutions. It is true that there have been advances in taking precise locations over the last several decades, but a physical location is better known if it is additionally described in archived field notes. Photographs of sites, taken at systematic directions or angles, can also help define locations (and their temporal condition) in ways that coordinates cannot. Considering repeatability can also help define what data to collect. While a highly specialized tool may collect useful data today, it is worth considering the potential lifespan of that tool and how easily the same or comparable data can be collected in the future.
A second lesson from the past is the importance of sampling gradients of variation. If the point of temporal analysis is to look for change, then sampling along gradients provides more opportunities for change as well as a fuller picture of change. The vast majority of published studies of climate change impacts have focused on one of three gradients: Shifts in phenological events (i.e., a seasonal gradient), shifts in elevational range, or shifts in latitudinal range. For those collecting data, the lesson here is that sampling only one relatively homogenous resource may not provide as much insight into the future as sampling a gradient of resources. Broadening a sampling scheme to include variation might require sampling resources that are not as ecologically intact, such as buffers or areas with greater human impacts. The value added through the transformation of inference on a single resource to inference on a gradient of a resource may strongly contribute to understanding the dynamic impacts of future environmental change (Hargrove and Rotenberry 2011).
Finally, the clearest lesson from the past is the importance of information access. Data may have no value in the future if they cannot be found, accessed, and understood. Ironically, while technology can make archiving easier, it can also make access harder. Data from the past that were stored on paper and securely archived are much easier to access today than data from the 1970s and 1980s that are archived on now-archaic data storage devices. The internet and the development of global data depositories appear to provide a potential solution as online databases may be transitioned into new forms as technology changes (see Box 13.2). Countries in Asia and Europe have a head start on the United States as observational and phenological databases are often organized nationally and controlled by the government (Primack and Miller-Rushing 2012). Nevertheless, the past warrants caution. Multiple forms of data storage are recommended (both online and offline), and care should be made to consider storing data in ways that will be most accessible and understandable to future researchers. Even with archiving, metadata are almost as important as the occurrence data itself.
Altogether, the past has lots to offer: It can inform us about the present and it may help prepare us for the future. But dealing with the past can be difficult and the rewards of digging up historical records and struggling to use them are not always immediately obvious. Hopefully this chapter can work as both a tour guide and self-help manual for the exploration of historical data and the facilitation of discovery. Historical data may only be one tool in the process to understand long-term environmental change, but it is a tool that has much to offer.
ACKNOWLEDGMENTS
Kim Hall, Lori Hargrove, Mark Herzog, Chrissy Howell, and Terry Root provided helpful comments and thoughtful advice on this chapter. Many of the ideas presented here were developed in conversations with Steve Beissinger, Craig Moritz, Jim Patton, and members of the Grinnell Resurvey Project at the University of California, Berkeley.
LITERATURE CITED
Bakker, J. D. and M. M. Moore. 2007. Controls on vegetation structure in southwestern ponderosa pine forests, 1941 and 2004. Ecology 88:2305–2319.
Beever, E. A., P. F. Brussard, and J. Berger. 2003. Patterns of apparent extirpation among isolated populations of pikas (Ochotona princeps) in the Great Basin. Journal of Mammalogy 84:37–54.
Blalock-Herod, H. N., J. J. Herod, and J. D. Williams. 2002. Evaluation of conservation status, distribution, and reproductive characteristics of an endemic Gulf Coast freshwater mussel, Lampsilis australis (Bivalvia: Unionidae). Biodiversity and Conservation 11:1877–1887.
Bradford, D. F., J. R. Jaeger, and S. A. Shanahan. 2005. Distributional changes and populations status of amphibians in the eastern Mojave Desert. Western North American Naturalist 65:462–472.
Brusca, R. C., J. F. Wiens, W. M. Meyer, J. Eble, K. Franklin, J. T. Overpeck, and W. Moore. 2013. Dramatic response to climate change in the Southwest: Robert Whittaker’s 1963 Arizona Mountain plant transect revisited. Ecology and Evolution 3:3307–3319.
Chen, I.-C., J. K. Hill, R. Ohlemüller, D. B. Roy, and C. D. Thomas. 2011. Rapid range shifts of species associated with high levels of climate warming. Science 333:1024–1026.
Colwell, M. A., T. Danufsky, R. L. Mathis, and S. W. Harris. 2002. Historical changes in the abundance and distribution of the American Avocet at the northern limit of its winter range. Western Birds 32:1–15.
Comte, L. and G. Grenouillet. 2013. Species distribution modelling and imperfect detection: Comparing occupancy versus consensus methods. Diversity and Distributions 19:996–1007.
Crimmins, S. M., S. Z. Dobrowski, J. A. Greenberg, J. T. Abatzoglou, and A. R. Mynsberge. 2011. Changes in climatic water balance drive downhill shifts in plant species’ optimum elevations. Science 331:324–327.
Daniels, R. A. 1998. Changes in the distribution of stream-dwelling crayfishes in the Schoharie Creek system, eastern New York State. Northeastern Naturalist 5:231–248.
Davidson, C. 2004. Declining downwind: Amphibian population declines in California and historical pesticide use. Ecological Applications 14:1892–1902.
Davidson, C., H. Shaffer, and M. Jennings. 2001. Declines of the California red-legged frog: Climate, UV-B, habitat, and pesticides hypotheses. Ecological Applications 11:464–479.
Erb, L. P., C. Ray, and R. Guralnick. 2011. On the generality of a climate-mediated shift in the distribution of the American pika (Ochotona princeps). Ecology 92:1730–1735.
Forister, M. L., A. C. McCall, N. J. Sanders, J. A. Fordyce, J. H. Thorne, J. O’Brien, D. P. Waetjen, and A. M. Shapiro. 2010. Compounded effects of climate change and habitat alteration shift patterns of butterfly diversity. Proceedings of the National Academy of Sciences of the United States of America 107:2088–2092.
Franklin, J., C. L. Coulter, and S. J. Rey. 2004. Change over 70 years in a southern California chaparral community related to fire history. Journal of Vegetation Science 15:701–710.
Graham, C. H., J. Elith, R. J. Hijmans, A. Guisan, A. T. Peterson, B. A. Loiselle, and The Nceas Predicting Species Distributions Working Group. 2008. The influence of spatial errors in species occurrence data used in distribution models. Journal of Applied Ecology 45:239–247.
Graham, C. H., S. Ferrier, F. Huettman, C. Moritz, and A. T. Peterson. 2004. New developments in museum-based informatics and applications in biodiversity analysis. Trends in Ecology & Evolution 19:497–503.
Hargrove, L. and J. T. Rotenberry. 2011. Spatial structure and dynamics of breeding bird populations at a distribution margin, southern California. Journal of Biogeography 38:1708–1716.
Harrison, S. H., E. I. Damschen, and J. B. Grace. 2010. Ecological contingency in the effects of climatic warming on forest herb communities. Proceedings of the National Academy of Sciences of the United States of America 107:19362–19367.
Kelly, A. E. and M. L. Goulden. 2008. Rapid shifts in plant distribution with recent climate change. Proceedings of the National Academy of Sciences of the United States of America 105:11823–11826.
Kopp, C. W. and E. E. Cleland. 2013. Shifts in plant species elevational range limits and abundances observed over nearly five decades in a western North America mountain range. Journal of Vegetation Science 25:135–146.
Labay, B., A. E. Cohen, B. Sissel, D. A. Hendrickson, F. D. Martin, and S. Sarkar. 2011. Assessing historical fish community composition using surveys, historical collection data, and species distribution models. PLOS ONE 6:e25145.
Larrucea, E. and P. Brussard. 2008. Shift in location of pygmy rabbit (Brachylagus idahoensis) habitat in response to changing environments. Journal of Arid Environments 72:1636–1643.
Lutz, J. A., J. W. van Wagtendonk, and J. F. Franklin. 2009. Twentieth-century decline of large-diameter trees in Yosemite National Park, California, USA. Forest Ecology and Management 257:2296–2307.
MacKenzie, D. I., J. D. Nichols, J. A. Royle, K. H. Pollock, L. L. Bailey, and J. E. Hines. 2006. Occupancy Estimation and Modeling. Academic Press, Burlington, MA.
Martin, K., G. A. Brown, and J. R. Young. 2004. The historic and current distribution of the Vancouver Island White-tailed Ptarmigan (Lagopus leucurus saxatllis). Journal of Field Ornithology 75:239–256.
Mclaughlin, B. C. and E. S. Zavaleta. 2013. Regional and temporal patterns of natural recruitment in a California endemic oak and a possible “research reserve effect.” Diversity and Distributions 19:1440–1449.
Miller, B. P., N. J. Enright, and B. B. Lamont. 2007. Record error and range contraction, real and imagined, in the restricted shrub Banksia hookeriana in south-western Australia. Diversity and Distributions 13:406–417.
Morelli, T. L., A. B. Smith, C. R. Kastely, I. Mastroserio, C. Moritz, and S. R. Beissinger. 2012. Anthropogenic refugia ameliorate the severe climate-related decline of a montane mammal along its trailing edge. Proceedings of the Royal Society B: Biological Sciences 279:4279–4286.
Moriarty, K. M., W. J. Zielinski, and E. D. Forsman. 2011. Decline in American Marten occupancy rates at Sagehen Experimental Forest, California. Journal of Wildlife Management 75:1774–1787.
Moritz, C., J. L. Patton, C. J. Conroy, J. L. Parra, G. C. White, and S. R. Beissinger. 2008. Impact of a century of climate change on small-mammal communities in Yosemite National Park, USA. Science 322:261–264.
Primack, R. B. and A. J. Miller-Rushing. 2012. Uncovering, collecting, and analyzing records to investigate the ecological impacts of climate change: A template from Thoreau’s Concord. Bioscience 62:170–181.
Primack, R. B., A. J. Miller-Rushing, and K. Dharaneeswaran. 2009. Changes in the flora of Thoreau’s Concord. Biological Conservation 142(3):500–508.
Rickart, E. A., S. L. Robson, and L. R. Heaney. 2008. Mammals of Great Basin National Park, Nevada: Comparative field surveys and assessment of faunal change. Monographs of the Western North American Naturalist 4:77–114.
Root, T. L., J. T. Price, K. R. Hall, S. H. Schneider, C. Rosenzweig, and J. A. Pounds. 2003. Fingerprints of global warming on wild animals and plants. Nature 421:57–60.
Rowe, R. 2007. Legacies of land use and recent climatic change: The small mammal fauna in the mountains of Utah. American Naturalist 170:242–257.
Rowe, R. J., J. A. Finarelli, and E. A. Rickart. 2010. Range dynamics of small mammals along an elevational gradient over an 80-year interval. Global Change Biology 16:2930–2943.
Rowe, R. J. and S. Lidgard. 2009. Elevational gradients and species richness: Do methods change pattern perception? Global Ecology and Biogeography 18:163–177.
Rowe, R. J., R. C. Terry, and E. A. Rickart. 2011. Environmental change and declining resource availability for small-mammal communities in the Great Basin. Ecology 92:1366–1375.
Rowlands, P. G. and N. J. Brian. 2001. Fishtail mesa: A vegetation resurvey of a relict area in Grand Canyon National Park, Arizona. Western North American Naturalist 61:159–181.
Rubidge, E. M., W. B. Monahan, J. L. Parra, S. E. Cameron, and J. S. Brashares. 2011. The role of climate, habitat, and species co-occurrence as drivers of change in small mammal distributions over the past century. Global Change Biology 17:696–708.
Shaffer, H. B., R. N. Fisher, and C. Davidson. 1998. The role of natural history collections in documenting species declines. Trends in Ecology & Evolution 13:27–30.
Shavit, A. and J. Griesemer. 2009. There and back again, or the problem of locality in biodiversity surveys. Philosophy of Science 76:273–294.
Shultz, A. J., M. W. Tingley, and R. C. K. Bowie. 2012. A century of avian community turnover in an urban green space in northern California. The Condor 114:258–267.
Thorne, J. H., B. J. Morgan, and J. A. Kennedy. 2008. Vegetation change over sixty years in the central Sierra Nevada, California, USA. Madroño 55:225–239.
Tingley, M. W. and S. R. Beissinger. 2009. Detecting range shifts from historical species occurrences: New perspectives on old data. Trends in Ecology & Evolution 24:625–633.
Tingley, M. W. and S. R. Beissinger. 2013. Cryptic loss of montane avian richness and high community turnover over 100 years. Ecology 94:598–609.
Tingley, M. W., M. S. Koo, C. Moritz, A. C. Rush, and S. R. Beissinger. 2012. The push and pull of climate change causes heterogeneous shifts in avian elevational ranges. Global Change Biology 18:3279–3290.
West, N. E. and T. P. Yorks. 2006. Long-term interactions of climate, productivity, species richness, and growth form in relictual sagebrush steppe plant communities. Western North American Naturalist 66:502–526.
Winker, K., D. D. Gibson, A. L. Sowls, B. E. Lawhead, P. D. Martin, E. P. Hoberg, and D. Causey. 2002. The birds of St. Matthew Island, Bering Sea. Wilson Bulletin 114:491–509.
Yang, D.-S., C. J. Conroy, and C. Moritz. 2011. Contrasting responses of Peromyscus mice of Yosemite National Park to recent climate change. Global Change Biology 17:2559–2566.